Simplified molecular input lineentry system based quantitative structure activity relationship QSAR Models

Abstract

Predictions of different molecular properties, Physical and chemical properties have newlinebeen a long studied problem in the field of organic, pharmacy, drug discovery and polymer. newlineHence a computational screening process arises to determine the said properties for newlinecandidate materials. So more powerful computers can determine the properties where large newlineamount of experimental and predicted data about the structure of different molecules are newlineavailable theoretically. This theoretical approach has actually proved to be beneficent in newlinechemistry and other branches of science, where the experimental analysis and synthesis is newlinetime consuming, laborious, expensive or even hazardous. Hence high quality data are newlinerequired with relevant molecular descriptors that can produce the models that try to take in newlineaccount the vulnerable factors which can affect the physico chemical properties. newlineQSPR studies attempt to predict the activity of tested compounds and suggest the newlinestructural features which could enhance the biological activity in the process of drug design. newlineQSAR helps to find and predict rate constants of numerous untested micro pollutants. Again newlinethis research deals with predicting the solubility of compounds using their physico chemical newlineproperties. SMILES structures have been chosen to train the data. QSPR models have been newlineused to study the relationship between the structure and parameter of solubility. Various newlinemachine learning was taken and ensembled. The results show that, ensemble approaches can newlinebe successfully used for prediction the solubility. The SMILES-based QSAR models built by newlinethe Monte Carlo optimization process are efficient enough to predict the divergent properties newlinesuch as (i) pIC50 for the inhibition of MNK1 (ii) Adsorption energy for polypropylene newlinepolymerization (iii) Inhibiton constant (-log Ki) for the serotonin 3(5-HT3) receptor, (iv) newlineSolubility of CO2 and N2 in different polymers, (v) Catalytic activities of ZN catalyst and newline(vi) SADT calculation. The described methodology is universal for situations where the aim newlineis to predict the response of an eclectic system to a variety of physicochemical and/or newlinebiochemical conditions. newline